A card is one chart, table, funnel, or number on a dashboard, and each one answers a single question like "how many users started this experience?" You pick what the card measures, how it's drawn, and optionally a target to hold it against.

Prerequisites

• A dashboard to put the card on. See Analytics dashboards overview.
• The events you want to measure already in Appcues. Events your product sends count only from the moment you start sending them. A Click to Track event created today already covers past activity, because Appcues records interactions in the background. See Raw Event Recording.

Open the card builder

Open a dashboard and click Add card. The card builder opens.

Under What do you want to measure?, choose a card type:

• Metrics — track a measurement, drawn as a trend over time, a breakdown, or a single value.
• Funnel — measure conversion across a sequence of steps within a conversion window.
• Retention — measure how many users come back after a starting event, grouped into cohorts.

Configure the card, then click Save card. The card is added to your dashboard.

Turn on Compare to break a Metrics card into a line or bar per group — segment, account, campaign, tactic, or a dimension such as page or platform — or to compare it against the previous period. Funnel and Retention cards don't have it. See Compare metrics and measure uplift.

You can also describe the card you want to Appcues AI instead of filling in the builder. See Analytics dashboards overview.

Metrics cards

A Metrics card tracks one measurement, shown as a chart, a table, or a single large number.

1. Choose a Metric, for example Unique count of Users.
2. Turn on Events and click Add event to choose the event the metric applies to. Most metrics are a complete question on their own, so Events is off by default, and narrows the metric once you turn it on. Turning it back off clears the events you picked. Conversion rate has no switch, because its events are the measurement.
3. Add a second event to chart both, then set Match to decide how the card reads them. The legend lists what's on the card and appears only when the card charts more than one line or bar, with names shown in sentence case.
4. At the top right of the preview, choose how to analyze the data: Trend over time, Breakdown, or Single value. Then pick a chart style from the buttons underneath.
5. Click Save card.

Match multiple events

With two or more events on a card, Match decides how the card reads them.

• Each — one line or bar per event, drawn side by side. Picking several events has always done this.
• All — a single line or bar, counting only the users, accounts, or sessions that did every event in the date range.
• Any — a single line or bar, counting whoever did at least one of them.

Match stays greyed out until the card names two events. With one event, all three readings are the same event.

Who All and Any count follows the metric. On Unique count of Users, All means the same user did every event. On Count of Accounts, anyone in the account did each one. Count of Events counts how many times the listed events happened among the users who did all of them, so the number is a count of events, not of users.

Switching to a metric that doesn't offer your current choice moves the card to one it does offer.

Choose how the card is drawn

Two controls decide how a Metrics card is drawn. The dropdown on the preview's title row picks between Trend over time, Breakdown, and Single value. The row of buttons directly below it picks the style within that choice.

The first style listed for each is its default. Changing the dropdown resets the style buttons to that default, so set the dropdown first and the style second.

A few things worth knowing:

• A Breakdown needs more than one bar — otherwise it draws a single bar for the whole metric. Add a second event under Events with Match set to Each, or turn on Compare and add two or more segments, accounts, campaigns, or tactics.
• Bars change meaning with the chart type. Under Trend over time the bars keep the time axis, one bar per period. Under Breakdown there is no time axis, and the bars compare groups instead.
• Bars and Stacked bars draw the same chart when there's only one thing to plot — stacking appears once the card has more than one line or bar.

Which comparisons you can add also depends on the chart type. See Compare metrics and measure uplift.

Show one number with Single value

Single value replaces the chart with one number. Pick which period it covers:

• Total period (default) — the total across the dashboard's whole date range. On a Last 30 days dashboard, an Onboarding completions card sums all 30 days into one number.
• Most recent — the value for the latest period, matching the dashboard's granularity. At daily granularity you get the most recent day's number; at weekly, the most recent week's; at monthly, the most recent month's. Change the granularity in the Date is filter and the number updates to match.

Funnel cards

A Funnel card shows how many users move through a sequence of events and where they drop off.

1. Set the Conversion time window, how long a user has to complete the funnel after entering it. Enter a number, then pick days, hours, minutes, or seconds. It defaults to 14 days.
2. Under Funnel events, click Add event inside Step 1 to choose the event that puts a user into the funnel, then do the same for each later step.
3. Click Add step to extend the sequence, or the × beside a step to remove it.
4. Preview the funnel as you add steps. Until the first event is added you'll see the prompt "Add at least one event to preview the funnel".
5. Click Save card.

A step can hold more than one event. Each step reads did any of, so a user who did any event in that step has cleared it. Group equivalent actions this way, for example every route your product offers for creating an item as a single first step.

Set the conversion window to match how long users realistically take. A funnel from signup to activation with a 1-day window will read as a drop-off problem when it is really a window problem.

Retention cards

A Retention card groups users into cohorts, then shows how many of them came back and did a second event in the periods after. Use it to tell a feature that users try once from a feature that becomes a habit.

Set the Start and Return events

Under Event, two slots appear in order, each defaulting to + (Any event):

1. Start — the event that puts a user into a cohort and marks period 0 for them.
2. Return — the event that counts as coming back.

The two can be the same event. Setting both to your feature's usage event answers "of the users who used this, how many used it again?" Setting Start to a signup event and Return to a feature event answers "of the users who signed up, how many reached this feature?"

Choose how to define the cohort

Cohort definition decides what one row of the grid is:

• Start event date (default) — one row per period, holding the users whose Start event fell in it.
• Segments — one row per segment you pick. A Select segments dropdown appears below.
• Accounts — one row per account you pick. A Select accounts dropdown appears below.

Both dropdowns are searchable and take up to 10 cohorts; an eleventh selection is ignored without a message. The card won't preview until you pick at least one.

Cohort period is read only and follows the dashboard's granularity. Change the granularity in the dashboard's Date is filter to change it.

Read the cohort grid

Retention cards have their own selector above the card, separate from the two controls on Metrics cards:

• Cohort grid — a table, one row per cohort.
• Trend over time — the same data as a single retention curve.

The grid has a Cohort column naming the row, a Size column for how many users are in it, then one column per period since that cohort's Start event. The letter follows the dashboard's granularity, so the columns run D0, D1, D2 on a daily dashboard, W0, W1, W2 on a weekly one, and M0, M1, M2 on a monthly one. D0 is the Start period itself. A period with no Start events gets no row, so the Cohort column can skip dates. Scroll sideways for the later periods.

Each cell is the share of that cohort's Size that did the Return event during that period, not a running total, so the numbers rise as well as fall as you read across.

The first column reads 100% whenever the Start event guarantees the Return event, which is what happens when both slots hold the same event. Leaving Return on (Any event) counts any page view, named "Any page view" in the card's subtitle, so it reads 100% for a Start event that arrives alongside a page view and lower for one that can arrive without one, such as an event tracked server-side through the API. Pick a Return event the Start event doesn't guarantee and D0 becomes meaningful.

Read one row and one column

A row is one cohort over time. Jul W1 '26, outlined below, is 60 users who added a block that week. W1 reads 40%, so 24 of them added another the following week. The row settles into the high teens and low twenties from W3 on, and 13% were still adding blocks seven weeks in. That flattening is the signal worth having.

A column is every cohort at the same age, which makes it the fair comparison. The W1 column, outlined below, reads 40%, 26%, 12%, 13%, 20%, 29% and 31% from Jul W1 down to Aug W3, and each of those describes a cohort one week after its own start.

Why the grid is triangular

Each row's columns count forward from that cohort's own Start date, so no two rows cover the same calendar dates. On a daily card, D3 on an Aug 2 cohort is Aug 5, while D3 on an Aug 20 cohort is Aug 23. The cohorts near the bottom started recently, so their later periods haven't happened yet and stay blank, one more blank per row as you go down.

A row stops where today is, not where the dashboard's date range ends. The range only decides whose Start event puts them in a cohort. After that, Appcues keeps counting their Return events up to today, past the end of the range if the range closed earlier.

What not to read into the grid

Three things in the grid look like findings and aren't.

• The last number in each row. It sits in the period the card is currently in, so it counts part of a period against the whole cohort and reads low. Nearly every row above ends at 0% for that reason. Read the second-to-last cell for the most recent complete period.
• The staircase edge read as a trend. Each cell along it is a different age, so following it down compares a cohort eight weeks old against one two weeks old.
• A percentage in a small cohort. The Jun W5 '26 row holds 11 users, partly because its week begins before the date range does and only the days inside the range count. Its climb from 27% at W3 to 55% at W4 is three more people, not a turnaround.

What the cell colors mean

Cells are color coded in four bands: dark green at 75% and above, green from 50% to 74%, amber from 25% to 49%, and red below 25%.

How red the grid looks depends on the Return event. A specific event, such as the same feature in both slots, usually retains below 25% after the first period or two, so most cells fall in the red band and that is normal rather than a sign the card is misconfigured. Leaving Return on (Any event) measures any return visit instead, which retains much higher and produces a mostly green grid.

Segment and account grids

When you cohort by segments or accounts the grid isn't triangular, but rows can still end at different columns. Each row runs as far as the periods its own users have lived through, so a segment whose users all started recently ends sooner than one with longer-standing users.

Trend over time

Trend over time plots retention against the period offset rather than the calendar, so the x-axis runs D0, D1, D2 and the line shows the shape of the decay. Use it when you care about how quickly retention falls off rather than about individual cohorts.

Add a target to a card

Targets overlay a goal line on a Metrics card so you can see actual performance against where you want it to be. Funnel and Retention cards don't have them.

1. In the card builder, turn on the Target toggle.
2. Choose Achieve to reach a value, or Maintain to hold one.
3. Choose At least or At most to set the direction.
4. Enter the target value. The unit matches your metric, for example users.
5. Click Save card, or Update card when editing an existing card. The target line appears on the chart.

The goal line follows the value axis, so it stays readable when you switch a card between vertical and horizontal bars.

If a card has no data

Card-specific checks first:

• Funnel — the conversion time window may be shorter than users realistically take. Widen it and see whether the later steps fill in.
• Retention — if only the leftmost columns have numbers, the cohorts are too recent for the later periods. Widen the dashboard's date range. An empty Retention card also names the reason: "No cohorts started in this date range" when you cohort by Start event date and nobody did the Start event, and "No retention activity in this date range" when you cohort by segments or accounts, where the cohorts exist but nobody returned.
• Metrics — a Conversion rate with a second event is measuring against that narrower group, so an empty result can mean the denominator is empty rather than the numerator.
• Metrics drawn as a Table — an em dash means that period was never measured, which is different from a measured zero. Widen the date range if a whole column reads as dashes.

For the checks that apply to every card, such as the date range, event filters and dashboard-level segment filters, see Analytics dashboards overview.